Start with one request.

The base URL, model ID, and secret key are the only changes your first integration needs.

export LEXELANE_API_KEY="your-secret-key"
# Keep this secret on your server.

from openai import OpenAI
import os

client = OpenAI(
    base_url="https://lexelane.com/v1",
    api_key=os.environ["LEXELANE_API_KEY"],
)

response = client.chat.completions.create(
    model="deepseek/deepseek-v4.1-flash",
    messages=[{"role": "user", "content": "Hello, Lexelane."}],
    max_tokens=256,
)
print(response.choices[0].message.content)

Available model IDs are returned by GET /v1/models with your API key. Coming-soon catalog entries cannot be called yet.

A key per application.

Sign in with an operator-provisioned email/password account, or Google when enabled. Password signup is not public. After signing in, Account settings lets you rotate your password; application API keys are created and revoked separately.

Use Authorization: Bearer with your secret key. The full secret is displayed once; the server stores its hash. Set a lifetime spending budget and expiry for each key, and revoke old keys from the workspace.

curl "https://lexelane.com/v1/models" \
  -H "Authorization: Bearer $LEXELANE_API_KEY"

Never embed secrets in browser bundles, URLs, screenshots, or a public repository.

Streaming & tool calls

Chat Completions uses server-sent events. Ask for final usage so the request can be accurately reconciled. Tool definitions and results use the selected model’s native schema; your application executes client-side tools.

stream = client.chat.completions.create(
    model="deepseek/deepseek-v4.1-flash",
    messages=[{"role": "user", "content": "Hello"}],
    max_tokens=256,
    stream=True,
    stream_options={"include_usage": True},
)
for chunk in stream:
    if chunk.choices:
        print(chunk.choices[0].delta.content or "", end="")

One service. Your existing tools.

model = "your-responses-model-id"
model_provider = "lexelane"

[model_providers.lexelane]
name = "Lexelane"
base_url = "https://lexelane.com/v1"
env_key = "LEXELANE_API_KEY"
wire_api = "responses"
requires_openai_auth = false
supports_websockets = false

export ANTHROPIC_BASE_URL="https://lexelane.com"
export ANTHROPIC_AUTH_TOKEN="$LEXELANE_API_KEY"
export ANTHROPIC_MODEL="your-messages-model-id"
export DISABLE_PROMPT_CACHING="1"
claude

export LEXELANE_BASE_URL="https://lexelane.com"
node packages/cli/src/cli.mjs models
node packages/cli/src/cli.mjs balance
node packages/cli/src/cli.mjs chat "deepseek/deepseek-v4.1-flash" "Hello"

{
  "mcpServers": {
    "lexelane": {
      "command": "node",
      "args": ["/absolute/path/to/lexelane-client-kit/packages/mcp/src/server.mjs"],
      "env": {
        "LEXELANE_BASE_URL": "https://lexelane.com",
        "LEXELANE_API_KEY": "your-secret-key"
      }
    }
  }
}

Predictable limits. Traceable requests.

Requests reserve a conservative budget before generation, then settle against confirmed usage. Missing final usage is flagged for review and may temporarily charge the reserved ceiling; confirmed overcharges are refunded through the ledger. Default account limits are 60 requests per minute and 5 concurrent calls; your workspace shows your actual limits.

Keep the x-request-id response header when contacting support.